The present invention relates generally to the field of resource placement in a cloud computing environment, and more particularly to optimizing a placement plan for a host failure in a cloud computing environment.
A cloud environment is generally composed of a collection of hosts, with virtual machines (VMs) created on top of the hosts. To ensure the cloud environment remains in an optimally balanced state, calculations are performed, often by a placement engine, in order to determine an optimal placement plan for the various components within the cloud environment. The calculations are partially based on received advice, or constraints, from one or more advisors. In response to a failure occurring, some systems may calculate a new placement plan to migrate VMs, partially calculated from the received advice, to account for the host failure. To further ensure the cloud environment continues to run in an optimized state, some systems store a recovery placement plan, which may be accessed at the time of a system failure.